Clinical Predictors of Good/Poor Response to Growth Hormone Treatment in Children with Idiopathic Short Stature

Abstract

Introduction

Children with idiopathic short stature (ISS) are known to have varying responses to growth hormone (GH) treatment (GHT).

Methods

We conducted a post hoc analysis to identify clinical characteristics predictive of good and poor response during year 1 of GHT. Data from the NordiNet^® IOS (NCT00960128) and the ANSWER Program (NCT01009905) were used. Patients were grouped according to their response to GHT; good, middle, and poor responders had a change in HSDS of >1.0, 0.4–1.0, and <0.4, respectively. Patients were also grouped according to their responsiveness to GH dose. Logistic regression modelling was performed to identify clinical characteristics predictive of response to GHT.

Results

The response analysis set included 207 patients. Patients were 3–11 years old (males) or 3–10 years old (females) at the start of treatment and were prepubertal throughout year 1 of treatment. Age at the start of treatment (odds ratio [OR] 0.69, 95% confidence interval [CI] 0.5; 0.9, p = 0.0169) and distance from target HSDS (OR: 2.05, 95% CI: 1.1; 3.9, p = 0.0259) were found to be significant independent predictors of being in the good- versus poor-response groups. When patients were grouped according to their responsiveness to GH dose, a positive correlation between GH dose and change in HSDS was observed.

Conclusion

We identified younger age and further distance from target HSDS as clinical predictors of response to GHT in a heterogenous population of children with ISS. Future studies, designed to identify the genetic determinants of response to GHT, could further facilitate individualisation of GHT.

Plain Language Summary

Idiopathic short stature (ISS) is the name for when a child is much shorter in height than expected for their age. Children with ISS may grow taller when they receive growth hormone treatment (GHT), but not always. We looked at data from children with ISS who had received GHT to identify factors that could predict how much a child grew on treatment. We used information from two reports where children with ISS were treated with GHT. Our study included 207 boys and girls with ISS who were aged between 3 and 11 years when they started GHT. None of the children had started puberty. We looked at how many children grew after 1 year on GHT compared with their expected height for their age and sex, termed response to GHT. Based on their response, children were defined as having a “good response,” “middle response,” or “poor response”. Then, we checked if any other factors affected how well children responded to GHT. Children with ISS who grew the most (had a good response to GHT) were younger at the start of treatment and further away from their expected height than children who grew the least (had a poor response to GHT). We also found that children who got a higher dose of GHT grew more than those who got a lower dose. These results may help doctors predict how much a child might grow on GHT, which could help doctors, patients and caregivers to make decisions about treatment.

Introduction

Idiopathic short stature (ISS) is a multifactorial disorder affecting a heterogenous population of children with short stature of unidentified causes [1–3]. Children with ISS are characterised as having a “height standard deviation score (HSDS) ≥2 SDS below the corresponding mean height for a given age, sex, and population, without evidence of systemic, endocrine, nutritional, or chromosomal abnormalities; they have a normal birth weight and length, and are growth hormone (GH) sufficient” [3]. Short stature is one of the most common concerns presenting to paediatric endocrinologists, and it is estimated that 60–80% of all short children with HSDS ≤–2 fit the definition of ISS [3]. However, due to the many potential causes of ISS, it is a diagnosis of exclusion [1].

GH treatment (GHT) is administered to children with ISS to achieve normal adult height [4], and several studies have shown that it is effective in achieving adult height within the expected range [5–8]. In a phase 2 trial investigating the efficacy and safety of GHT in children with ISS, GHT resulted in significant improvements in HSDS and other growth-related variables [8]. The study involved 360 children with ISS and the results showed that GHT was well tolerated. In a randomised controlled trial of children with ISS receiving one of two doses of GH (33 µg/kg/day, n = 18 or 67 µg/kg/day, n = 31, respectively) versus controls (n = 19), children who received GH achieved higher gains in HSDS than children in the control group (1.2 and 1.3 versus 0.4, respectively) [7]. In the same study, children with parental height in the expected range responded best to GHT.

Despite the evidence, the use of GHT in children with ISS is controversial and not yet approved in Europe [1, 9, 10]. This is due to several reasons, among which is the diagnosis of ISS being purely descriptive and the variable responsiveness of patients to GHT; the adult height of some patients with ISS treated with GHT remains below the expected range [11]. This is not surprising, as ISS encompasses many unidentified underlying causes [2].

To attempt to mitigate the variability of responsiveness to GHT in patients with ISS, the authors of one study developed and validated multivariate models to predict the response of children with short stature to GHT [12]. The study included data on children with isolated growth hormone deficiency (GHD) or ISS. The models were developed using five sets of clinical information: auxological data from the year prior to GHT start, growth data from the first 2 years of life, insulin-like growth factor-I (IGF-I) data, the results of a GH secretion estimation during a provocation test or a spontaneous GH secretion profile, and parental heights. The results of the study showed that the model provided an estimate of GH responsiveness in children with short stature with an accuracy that was better than a provocation test. In a similar study, a multiple regression analysis was conducted to determine the factors characterising the growth response of patients with ISS reaching near adult height (n = 256) using data from the Kabi Pharmacia International Growth Study (KIGS) database [13]. The results showed that the magnitude of growth response was negatively correlated with chronological age and HSDS minus mid-parental HSDS, but positively correlated with GH dose and weight at the start of GH treatment. The aim of both studies was to develop a model with practical utility.

Most children with ISS can attain adult height within the normal range without intervention [1]. Some are treated with GH to improve the growth rate; however, treatment outcomes vary and not all patients respond to GHT [1]. Determining the clinical characteristics that influence response to GHT in children with ISS may help predict their response to GHT. This, in turn, could facilitate individualisation of their treatment. This analysis of data from the NordiNet^® International Outcome Study (IOS) (NCT00960128) and the American Norditropin Studies: Web-Enabled Research (ANSWER) Program (NCT01009905) was designed to investigate the clinical characteristics that are associated with extremes of response to GHT during the first year of treatment, defined as good or poor responders, in children with ISS.

Methods

Study Design

This post hoc analysis was conducted using data from the NordiNet^® IOS and the ANSWER Program. The detailed study designs and methodologies have been previously published [14, 15]. In brief, NordiNet^® IOS and the ANSWER Program were two complementary, observational registry studies investigating the long-term effectiveness and safety of GHT as prescribed in routine clinical practice.

Both studies included paediatric patients with growth disorders receiving GHT. The subset of GH-naïve children with ISS, after exclusion of children born small for gestational age (SGA), was 1,499.

Both studies were approved by the relevant ethics committees, conducted with written consent from patients. Pseudonymisation of data was performed in accordance with the Declaration of Helsinki, regulatory requirements, and the Guideline for Good Pharmacoepidemiology Practices.

Patient Disposition

From the subset of children with ISS identified in the registry studies and after exclusion of children born SGA (birth length SDS <−2 or birth weight SDS <−2), the following inclusion criteria were applied: gestational age ≥32 weeks, prepubertal throughout the first year of GHT, age at the start of treatment 3–11 years for boys and 3–10 years for girls, no concomitant medication that can influence GHT, no concomitant illness that can influence GHT, HSDS at baseline measured <2 months prior to GHT start, and availability of data for HSDS change from baseline at year 1 (Fig. 1). For concomitant medications or illnesses that can influence GHT, the exclusion timeframe was from prior to initiation of GHT and up to the first year of treatment. These inclusion and exclusion criteria were chosen to ensure that the study sample was homogenous and the focus of the study was on the clinically relevant ISS patient population. The criteria were also set to mitigate confounding factors such as puberty.

Fig. 1.

Patient disposition, response analysis set. ^aEffectiveness analysis set from NordiNet^® IOS (NCT00960128) and the ANSWER Program (NCT01009905). ^bThe timeframe of receipt of any medicinal products or the presence of concomitant illnesses that could influence response to GHT was within 3 months before or during the first year of GHT (exception: ADHD medications were allowed). ADHD, attention-deficit hyperactivity disorder; GHT, growth hormone treatment; HSDS, height standard deviation score; ISS, idiopathic short stature; SGA, small for gestational age.

The response analysis set comprised patients with ISS who met the inclusion criteria. From the response analysis set, patients were grouped according to their response to GHT. Children with a change in HSDS of >1.0 in the first year of GHT, which corresponded to having a change in HSDS in the >85th percentile of the response analysis set, comprised the good-response group. Children with a change in HSDS of <0.4 in the first year of GHT, which corresponded to having a change in HSDS in the <15th percentile of the response analysis set, comprised the poor-response group. Children in the middle-response group had a change in HSDS of 0.4–1.0 in the first year of GHT.

Due to the known dose-dependent effect of GH, patients were also grouped according to their responsiveness to GH dose; the degree of response was adjusted for the dose of GH received. For this evaluation, children with high responsiveness to GH dose were defined as having a change in HSDS of >0.23 per 10 μg/kg/day GH. Children with low responsiveness to GH dose were defined as having a change in HSDS of <0.08 per 10 μg/kg/day GH. This again corresponded to children with a change in HSDS per 10 μg/kg/day GH in the >85th percentile or <15th percentile of the response analysis set, respectively. Children were classified as having medium responsiveness to GH dose if they had a change in HSDS of 0.08–0.23 per 10 μg/kg/day GH.

Responsiveness to GH dose was calculated as follows:

$$Responsiveness=\frac{Annualised∆HSDS}{Averagedoseinyear1}$$

Average dose in year 1 was computed as the weighted average of relative doses for the duration of each treatment period during year 1. This was done to mitigate instances of more than one prescription at year 1 of treatment.

Variables of Interest Associated with Response to GHT and Responsiveness to GH Dose

The following variables were identified as potential predictors of response to GHT: sex, gestational age (weeks), birth weight SDS, birth length SDS, target HSDS, mother’s HSDS, father’s HSDS, IGF-I at baseline, GH peak (ng/mL) at baseline, age at the start of treatment, body mass index (BMI) SDS at baseline, HSDS at baseline, distance from target HSDS at baseline (measured as target HSDS – HSDS at baseline), ratio of bone age to chronological age, dose at baseline (μg/kg/day), average dose at year 1 (µg/kg/day), GH dose (µg/kg/day) change from baseline at year 1 of treatment. The same aforementioned variables were selected for the analysis of responsiveness to GH dose.

Statistical Analyses

The analyses of individual variables to determine predictiveness of response to GHT were conducted using univariate logistic regression models. All identified variables of interest were selected for the univariate analyses.

The analyses of multiple variables predictive of response to GHT were conducted using multivariate logistic regression models. Variables selected for inclusion in multivariate analyses were sex, age at the start of treatment, and distance from target HSDS at baseline (measured as target HSDS – HSDS at baseline). Variable of average dose (10 µg/kg/day) at year 1 of treatment was included in the multivariate analyses of good versus poor response only. HSDS values included in the analyses were annualised per the equation:

$$Annualised∆HSDS=\frac{{HSDS}_1-{HSDS}_0}{t_1-{tgh}_{start}}$$

HSDS₁ = HSDS at year 1; HSDS₀ = HSDS at baseline; t₁ = time when HSDS close to 1 year was collected; and tgh_start = time when GH was started.

Results

Clinical Characteristics

In total, the response analysis set comprised 207 patients. The number of patients in the good- and poor-response groups was 31 for each, and the number of patients in the middle-response group was 145.

Clinical characteristics by responder groups are presented in Table 1. Female patients comprised 26%, 36%, and 16% of the poor-, middle-, and good-response groups, respectively. Mean age at the start of treatment was lowest for the good-response group (6.39 years) and highest for the middle-response group (8.10 years). Additionally, mean age at the start of treatment was higher in the poor-response group compared with the good-response group (7.73 vs. 6.39 years). Distance from target HSDS, measured as target HSDS minus HSDS at baseline, was highest for the good-response group (2.30), then the middle-response group (1.84), followed by the poor-response group (1.77). Average GH dose (μg/kg/day) at year 1 of treatment was higher for both the good- and the poor-response groups (49.16 per μg/kg/day and 49.05 per μg/kg/day, respectively) than the middle-response group (46.56 per μg/kg/day).

Table 1.

Individual variable statistics by poor-, middle-, and good-responder groups, and the odds ratio of having a good versus poor response, response analysis set

Variable	Responder group	N	Mean (SD)	Odds ratio (95% CI) of good vs. poor response; p value
Female patients, %	Poor	8 (26%)	–	0.55 (0.16; 1.93); 0.35291
	Middle	52 (36%)	–
	Good	5 (16%)	–
Gestational age, weeks	Poor	31	38.84 (2.1)	0.93 (0.73; 1.18); 0.52776
	Middle	145	39.08 (1.8)
	Good	31	38.50 (2.2)
Birth weight SDS	Poor	31	−0.45 (1.0)	1.03 (0.58; 1.84); 0.92051
	Middle	137	−0.34 (1.0)
	Good	30	−0.43 (0.8)
Birth length SDS	Poor	23	−0.29 (1.3)	1.02 (0.62; 1.69); 0.92687
	Middle	80	−0.56 (1.1)
	Good	17	−0.25 (1.3)
Target HSDS	Poor	30	−1.10 (0.7)	2.66 (1.20; 5.89); 0.01599
	Middle	133	−0.89 (0.8)
	Good	28	−0.60 (0.7)
Mother HSDS	Poor	30	−1.26 (0.8)	1.78 (0.98; 3.23); 0.05769
	Middle	137	−1.12 (1.2)
	Good	30	−0.78 (1.0)
Father HSDS	Poor	30	−0.97 (1.1)	1.47 (0.88; 2.46); 0.13885
	Middle	133	−0.70 (0.9)
	Good	28	−0.55 (1.0)
IGF-I SDS at baseline	Poor	12	−1.09 (1.2)	1.07 (0.62; 1.86); 0.79810
	Middle	81	−0.91 (1.2)
	Good	19	−0.96 (1.5)
GH peak at baseline, ng/mL	Poor	8	22.90 (17.5)	0.96 (0.89; 1.04); 0.32216
	Middle	43	15.54 (7.3)
	Good	14	17.61 (7.0)
Age at the start of treatment, years	Poor	31	7.73 (1.8)	0.72 (0.55; 0.94); 0.01482
	Middle	145	8.10 (1.8)
	Good	31	6.39 (2.2)
BMI SDS at baseline	Poor	31	−0.13 (1.2)	1.02 (0.67; 1.56); 0.93451
	Middle	145	−0.40 (1.0)
	Good	30	−0.10 (1.2)
HSDS at baseline	Poor	31	−2.87 (0.7)	0.80 (0.36; 1.79); 0.59225
	Middle	145	−2.71 (0.5)
	Good	31	−2.95 (0.6)
Distance from target HSDSa	Poor	30	1.77 (1.1)	1.72 (0.98; 3.00); 0.05739
	Middle	133	1.84 (0.9)
	Good	28	2.30 (0.9)
BA/CA at baseline	Poor	14	0.80 (0.2)	0.63 (0.01; 31.35); 0.8184
	Middle	82	0.84 (0.1)
	Good	17	0.78 (0.2)
Dose at baseline, µg/kg/day	Poor	31	46.15 (19.9)	1.00 (0.98; 1.03); 0.73597
	Middle	145	46.76 (12.8)
	Good	31	47.86 (20.5)
Average dose at year 1, µg/kg/day	Poor	31	49.2 (15.2)	1.00 (0.97; 1.04); 0.97595
	Middle	145	46.6 (11.9)
	Good	31	49.1 (14.0)
GH dose change from baseline at year 1, µg/kg/day	Poor	30	4.0 (18.8)	1.00 (0.97; 1.02); 0.79063
	Middle	133	0.3 (8.4)
	Good	28	2.5 (24.0)
HSDS change from baseline at year 1	Poor	31	0.26 (0.1)	–
	Middle	145	0.71 (0.2)
	Good	31	1.23 (0.2)

BA/CA, bone age/chronological age; BMI, body mass index; CI, confidence interval; GH, growth hormone; HSDS, height standard deviation score; IGF-I, insulin-like growth factor-I; N, number of patients; SD, standard deviation; SDS, standard deviation score. ^aMeasured as target HSDS − HSDS at baseline.

Univariate Analyses: Individual Variables Predictive of Good versus Poor Response

Univariate analyses of clinical determinants of the good- versus poor-response groups suggested age at the start of treatment (odds ratio [OR] 0.72, 95% confidence interval [CI] 0.55; 0.94, p = 0.01482) and target HSDS (OR: 2.66, 95% CI: 1.20; 5.89, p = 0.01599) to be significant predictors of being in the good- versus poor-response groups, when tested individually. Distance from target HSDS (OR: 1.72, 95% CI: 0.98; 3.00, p = 0.05739) and mother’s HSDS (OR: 1.78, 95% CI: 0.98; 3.23, p = 0.05796) were also found to be strong predictors of good versus poor response to GHT when tested individually, although not statistically significant (Table 1).

Multivariate Analyses: Multiple Predictors of Good versus Poor Response

The results of the multivariate logistic regression model showed that younger age at the start of treatment (OR: 0.69, 95% CI: 0.51; 0.94, p = 0.0169) and further distance from target HSDS (OR: 2.05, 95% CI: 1.09; 3.85, p = 0.0259) were independent predictors of being in the good-responder group (Table 2). Average dose (10 µg/kg/day) at year 1 of treatment (OR: 0.77, 95% CI: 0.50; 1.18, p = 0.2333) and sex (OR: 0.25, 95% CI: 0.05; 1.25, p = 0.0912) were not found to be significant independent predictors of response to GHT.

Table 2.

Multivariate logistic regression of good vs. poor response, response analysis set

Effect	OR estimate	95% CI	p value
Female vs. male	0.251	0.050; 1.248	0.0912
Age at the start of treatment	0.689	0.507; 0.935	0.0169
Distance from target HSDSa	2.050	1.090; 3.854	0.0259
Average dose at year 1 (10 µg/kg/day)	0.769	0.499; 1.184	0.2333

CI, confidence interval; HSDS, height standard deviation score; OR, odds ratio.

^aMeasured as target HSDS − HSDS at baseline.

Multiple Predictors of High versus Low Responsiveness

In a multivariate logistic regression model of high versus low responsiveness to GH dose, being female (OR: 0.24, 95% CI: 0.06; 0.95, p = 0.0421) was related to low responsiveness to GH dose while further distance from target HSDS (OR: 1.85, 95% CI: 1.02; 3.36, p = 0.0446) was shown to be a predictor of having high responsiveness to GH dose (Table 3). Age at the start of treatment was not identified as a predictor of responsiveness to GH dose.

Table 3.

Multivariate logistic regression of high vs. low responsiveness to GH dose, response analysis set

Effect	OR estimate	95% CI	p value
Female vs. male	0.239	0.060; 0.950	0.0421
Age at the start of treatment	0.871	0.648; 1.171	0.3607
Distance from target HSDSa	1.847	1.015; 3.361	0.0446

CI, confidence interval; GH, growth hormone; HSDS, height standard deviation score; OR, odds ratio.

^aMeasured as target HSDS − HSDS at baseline.

GH Dose and HSDS

No correlation was observed between the average dose of GH and change in HSDS after the first year of treatment (Fig. 2a). However, after grouping patients according to their responsiveness to GH, a positive correlation was observed within responsiveness groups between GH dose and change in HSDS after 1 year of treatment (Fig. 2b). This positive correlation was significant for the high (p = 0.003) and medium responsiveness (p < 0.001) groups.

Fig. 2.

Change from baseline in HSDS at year 1 of GHT plotted against average dose (µg/kg/day) of GHT in year 1 without grouping (a) and with responsiveness groups (b). The positive correlation was significant for the high (p = 0.003) and medium responsiveness (p < 0.001) groups. GHT, growth hormone treatment; HSDS, height standard deviation score; ISS, idiopathic short stature.

HSDS change from baseline to the last measurement recorded by GH dose responsiveness groups is shown in Figure 3. Children who achieved a HSDS increase of ≥2 after 3 years of treatment were significantly younger at the start of treatment compared with children who achieved a HSDS increase of <2 (6.83 vs. 8.13 years, p = 0.0026).

Fig. 3.

Change from baseline in HSDS until last measurement by responsiveness groups, response analysis set. a Low-responsiveness group (children with a change in HSDS of <0.08 per 10 μg/kg/day GH in the first year of GHT). b Medium-responsiveness group (children with a change in HSDS of 0.08–0.23 per 10 μg/kg/day GH in the first year of GHT). c High-responsiveness group (children with a change in HSDS of >0.23 per 10 μg/kg/day GH in the first year of GHT). Diamonds represent the timepoint at which females reached 10 years of age and males reached 11 years of age. GH, growth hormone; GHT, growth hormone treatment; HSDS, height standard deviation score.

The number of children who achieved a height within the normal range by the last recorded height measurement was 29 (94%) in the high-responsiveness group, 115 (79%) in the middle-responsiveness group, and 17 (55%) in the low-responsiveness group. In the same order, the last recorded height measurement was obtained after a median of 4.7, 3.2, and 2.8 years on treatment.

Discussion

The aim of this post hoc analysis was to identify clinical characteristics that can be used as predictors of good versus poor response in the first year of GHT, using data from a subset of children with ISS who participated in the NordiNet^® IOS and the ANSWER Program. The results showed that age at the start of treatment, target HSDS, distance from target HSDS, and mother’s HSDS were strong predictors of good versus poor response to GHT during the first year of treatment. Additionally, a positive correlation between GH dose and change in HSDS in children stratified by high, medium, or low responsiveness to GH dose was also observed. This could indicate that a change in HSDS in the first year of treatment may be a better predictor of achieving adult height if evaluated in combination with responsiveness to GH dose, as opposed to change in HSDS alone.

Age at the start of treatment has been previously investigated in relation to predicted growth response in patients with ISS reaching near adult height in a multiple regression analysis of data from the KIGS database [13]. The magnitude of growth response was shown to be negatively correlated with chronological age [13]. This is in line with the results of our analysis, where a higher proportion of children in the good-response group started treatment at a younger age than those in the poor- or middle-response groups.

In addition to age at the start of treatment, the correlation of growth response to GH dose was also investigated in the aforementioned analysis. The results showed that the magnitude of growth response was positively correlated with GH dose [13]. Our analysis showed that children in the good-response group received a similar average GH dose at the first year of treatment to children in the poor-response group. Additionally, both groups received an average GH dose at 1 year of treatment that was higher than that received by the middle-response group. A possible explanation to this observation is that the poor-response group received larger increases in dose to compensate for their initial poor response. This can be further discerned by the numerically higher mean change in GH dose from baseline to year 1 in the poor-response group versus the good-response group (4.0 vs. 2.5 µg/kg/day, respectively, p = 0.79 in the univariate analysis). Similarly, by responsiveness grouping, more dose fluctuations were observed for children in the low-responsiveness group in the first year of treatment compared with children in the medium- and high-responsiveness groups.

Research on GH dosing in children with ISS has previously shown that a uniform weight-based dose for children with ISS is inefficient, as it does not consider the inherent individual sensitivities to GH [2, 7, 8, 16]. In a 4-year, open-label trial, patients with ISS were randomised to an individualised formula-based dose of GH for 2 years (n = 202), followed by a dose reduction for 2 years, or to a standard weight-based dose for 4 years (n = 114) [16]. Both groups achieved similar average height gains; however, the group receiving the individualised GH dose utilised less GH. In another trial, children with ISS (n = 108) were randomised to one of two doses of GH (67 or 33 μg/kg/day) or to the control group [7]. Patients receiving the higher GH dose achieved a higher mean (standard deviation [SD]) final HSDS compared with patients receiving the lower dose (−1.4 [0.86] vs. −1.7 [0.70], p = 0.032). This dose-response relationship with GHT was observed after the first year of treatment and up until final height. A similar phase 2 study was conducted in children with ISS where children were randomised to a high dose (0.2 mg/kg/week), low dose (0.1 mg/kg/week), or control group for 52 weeks (n = 120 for each group) [8]. Annualised mean height velocity of the group receiving the higher dose of GHT was significantly higher than that of the group receiving the lower dose of GHT (p < 0.0001). The findings from the aforementioned studies indicate that a higher GH dose may be more effective for children with ISS to achieve normal height.

There is variability in the effect of GH dose on children with ISS, which may lie in the heterogeneity of the population of children with ISS. In this study, we have been able to show this heterogeneity by grouping patients based on their responsiveness to GH dose. When patients with similar responsiveness to GH dose were grouped together, we observed a positive correlation between GH dose and change in HSDS. Dose responsiveness during the first year of treatment was also shown to have a correlation with remaining on treatment and achieving normal height. Children in the high-responsiveness group were on treatment the longest (4.7 years vs. 3.2 and 2.8 for children in the medium- and low-responsiveness groups, respectively) and had the highest proportion of children who achieved normal height (94% vs. 79% and 55%, respectively). It was also observed in this study that changes in HSDS from baseline tend to reach a plateau after a few years of treatment. The timepoint at which these plateaus were reached seems to have varied per patient, as can most clearly be observed in the high-responsiveness group. This is likely to be due to completion of puberty but may also reflect the heterogeneity of patients with ISS and their responsiveness to GH dose. It can be observed from the figures that children with a HSDS that reached a plateau earlier were older at start of treatment compared with children who reached the plateau later in their treatment years. This is supported by the results of children achieving a HSDS increase of ≥2 after 3 years of treatment being significantly younger at the start of treatment than those who achieved a HSDS increase of <2 in the same time period (6.83 years for the former group and 8.13 years for the latter group, p = 0.0026).

It has previously been suggested that target height, using mid-parental height, provides an accurate estimation that can be used to evaluate response to GHT [17]; however, for children with ISS, it is known that, on average, the adult height achieved will be below the parental target height [3, 18]. This study showed that, in children with ISS, target HSDS was indeed a significant predictor of response to GHT. However, further distance from target HSDS and mother’s HSDS were also shown to be strong predictors of response to GHT. Children in the good-response group had the highest target HSDS (−0.6 vs. −0.89 and −1.10 for the middle- and poor-response groups, respectively) and had the highest numerical distance from target HSDS (2.3 vs. 1.84 and 1.77, respectively). The underlying aetiology of short stature in children with ISS is unknown, and this patient population includes individuals with a heterogenous genetic background. Identifying the genetic determinants of response to GHT could help identify patients who would benefit from treatment and could support in individualised GHT paradigms for children with ISS. Ultimately, studies that analyse final adult height are necessary to consolidate the predictive variables as influencers of the treatment.

The analyses presented in this study have some limitations. As with all registry studies, data relating to clinical determinants and individual factors associated with patient health were not always available from all patients. This could have limited the analytical power of this study. Genetic testing was not part of the two registry studies used in this analysis, and we could not assess the impact of individual genetic variability on the response to treatment. A larger study including genetic assessments would provide more clarity on the role of different clinical and genetic determinants of response to GHT in children with ISS. Data on final height of children in each response group were not available for the majority of patients; thus, data of last recorded measurement were included in the analyses instead. Changes in diagnostic and eligibility guidelines, as well as in criteria for GHT in ISS populations, could have occurred during the course of the registry studies.

Despite the limitations outlined above, this study compared the extreme ends of a response distribution curve to identify clinical predictors of response to GHT. The resulting predictors identified in this analysis are in line with predictors identified in studies that utilised a larger sample size. As timing of clinic visits were per routine clinical practice, HSDS at 1 year of the start of treatment was annualised to account for visits that did not occur exactly 12 months from baseline. The findings of this analysis support the design of an ongoing study by the authors (NCT05894876), in which a similar study approach is being utilised to identify the genetic predictors of having a good or poor response to GHT in prepubertal children with ISS using whole genome sequencing.

Conclusions

This study identified age at the start of treatment and distance from target HSDS to be significant independent predictors of having a good versus poor response to GHT during the first year of treatment in children with ISS. The differing levels of responsiveness of patients with ISS to GH dose indicate that a variable sensitivity to GHT exists in this population. This variability could be due to the genetic heterogeneity of the ISS population. Identifying the genetic determinants of response to GHT in patients with ISS could help in developing a personalised approach to treatment.

Acknowledgements

The authors thank the investigators and patients who participated in this study. Editorial support (in the form of writing assistance, collating author comments, assembling tables/figures, grammatical editing, and referencing) was provided by Emra Baker and Helen Marshall, of Ashfield MedComms, and was funded by Novo Nordisk Health Care AG.

Statement of Ethics

Approval was obtained from relevant ethics committees, written informed consent was obtained, and all data were anonymised. For paediatric patients, the parent(s) or a legally authorised representative granted consent. The data used in this study were obtained from the NordiNet^® IOS and ANSWER Program, which were conducted in accordance with the Declaration of Helsinki and approved by the local institutional ethics committee/institutional review board, and the local regulatory authorities at each study centre and data privacy agencies as required. NordiNet^® IOS and ANSWER were also conducted in accordance with the Good Pharmacoepidemiology Practice guidelines.

Conflict of Interest Statement

A.D. received consultancy and grant funding from Novo Nordisk, Pfizer, and BioMarin and received consultancy from Bridge Bio and Tyra. M.P. received consulting fees, honoraria, and research grants from Novo Nordisk. J.-M.F. is a consultant for Novo Nordisk. N.K. is an employee of Novo Nordisk and holds stocks in Novo Nordisk. N.N. is an employee of Novo Nordisk Health Care AG and a stockholder of Novo Nordisk A/S. A.H.O. is an employee of and shareholder in Novo Nordisk. A.A.L.J.: received consulting fees from Novo Nordisk, is the principal investigator of an independent study sponsored by BioMarin, and receives a research grant for the National Council for Scientific and Technological Development (303294/2020-5 CNPq).

Funding Sources

This work was funded by Novo Nordisk. The funding source was involved in decisions relating to the study design, execution and analysis, and manuscript conception, planning, writing, and decision to publish. NordiNet^® IOS is registered at ClinicalTrials.gov (NCT00960128). The ANSWER Program is registered at ClinicalTrials.gov (NCT01009905). Medical writing support for the development of this manuscript, under the direction of the authors, was provided by Emra Baker, of Ashfield MedComms, an Inizio Co., and funded by Novo Nordisk.

Author Contributions

Conceptualisation and writing – review and editing: A.D., M.P., J.-M.F., N.K., N.N., A.H.O., and A.A.L.J. Data curation and formal analysis: J.-M.F. Investigation: A.A.L.J. and N.N. Methodology: J.-M.F. and N.N. Software and supervision: A.A.L.J., M.P., and N.N. Project administration, funding acquisition, resources, validation, visualisation, and writing – original draft: N.N.

Funding Statement

This work was funded by Novo Nordisk. The funding source was involved in decisions relating to the study design, execution and analysis, and manuscript conception, planning, writing, and decision to publish. NordiNet^® IOS is registered at ClinicalTrials.gov (NCT00960128). The ANSWER Program is registered at ClinicalTrials.gov (NCT01009905). Medical writing support for the development of this manuscript, under the direction of the authors, was provided by Emra Baker, of Ashfield MedComms, an Inizio Co., and funded by Novo Nordisk.

Data Availability Statement

The data that support the findings of this study are not publicly available because they contain information that could compromise the privacy of research participants but are available from the corresponding author upon reasonable request.